js_regex 3.7.2 → 3.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3ec1645dcb85514957cbc0d1d08d973c44386e7ae554b83ceb56be807178a08d
-  data.tar.gz: cebf15dc34ca2e2f6ed1e57cc9a4719ee79c652070d8d0b687dae9278ed18e8d
+  metadata.gz: 8971658980813740deb03ece3c5af5bfdfd7412f0630fc4a3f172e4c06b11c52
+  data.tar.gz: bdafa3639a230b1ec1ac4661828050d99339eb18ec1768fa2a6f1b5e69d95f1b
 SHA512:
-  metadata.gz: 74eb573819cf814ca8c196dae4a6ddbb139b941947197cb115153ec749a89fe5189b1296693f86596ded4a882f81858b877b1514ff346289510dce351df8224f
-  data.tar.gz: 07e5743c335d15bcadfd465df576e32a5b9c6c27918b16c0282f1ac19ecf5ba7634fdf6b9a528963742316ce1a15f174098934da61d2c0a8975f639e1827a4b0
+  metadata.gz: 13abdf7b41485194f05cce79751ca60e8b1f9fc864b17f58294650df9f9e485a889b9571d847bf564aa12b709fd572c53f27a5a2900e3dc8bfa765f522b58e62
+  data.tar.gz: 31941c0d7a4842fdea84d5f649f3df30c54e8da24e6dbf722cff3e48661ae89646117f66f33ee419c0fc0e393b11f284cb565751b4f77aa5eb0bbbc8d38d903d

data/lib/js_regex/conversion.rb

@@ -1,8 +1,8 @@
 class JsRegex
   #
-  # This class acts as a facade, passing a regex to the converters.
+  # This class acts as a facade, passing a Regexp to the Converters.
   #
-  # ::of returns a source String, options String, and warnings Array.
+  # ::of returns a source String, options String, warnings Array, target String.
   #
   class Conversion
     require 'regexp_parser'
@@ -10,28 +10,33 @@ class JsRegex
     require_relative 'error'
     require_relative 'node'
     require_relative 'second_pass'
+    require_relative 'target'
 
     class << self
-      def of(input, options: nil)
-        source, warnings = convert_source(input)
-        options_string   = convert_options(input, options)
-        [source, options_string, warnings]
+      def of(input, options: nil, target: Target::ES2009)
+        target                       = Target.cast(target)
+        source, warnings, extra_opts = convert_source(input, target)
+        options_string               = convert_options(input, options, extra_opts)
+        [source, options_string, warnings, target]
       end
 
       private
 
-      def convert_source(input)
+      def convert_source(input, target)
         tree = Regexp::Parser.parse(input)
-        context = Converter::Context.new(case_insensitive_root: tree.i?)
+        context = Converter::Context.new(
+          case_insensitive_root: tree.i?,
+          target:                target,
+        )
         converted_tree = Converter.convert(tree, context)
         final_tree = SecondPass.call(converted_tree)
-        [final_tree.to_s, context.warnings]
+        [final_tree.to_s, context.warnings, context.required_options]
       rescue Regexp::Parser::Error => e
         raise e.extend(JsRegex::Error)
       end
 
-      def convert_options(input, custom_options)
-        options = custom_options.to_s.scan(/[gimuy]/)
+      def convert_options(input, custom_options, required_options)
+        options = custom_options.to_s.scan(/[gimsuy]/) + required_options
         if input.is_a?(Regexp) && (input.options & Regexp::IGNORECASE).nonzero?
           options << 'i'
         end

data/lib/js_regex/converter/anchor_converter.rb

@@ -13,17 +13,39 @@ class JsRegex
         when :bol, :bos then '^'
         when :eol, :eos then '$'
         when :eos_ob_eol then '(?=\n?$)'
-        when :word_boundary then pass_boundary_with_warning('\b')
-        when :nonword_boundary then pass_boundary_with_warning('\B')
+        when :word_boundary then convert_boundary
+        when :nonword_boundary then convert_nonboundary
         else
           warn_of_unsupported_feature
         end
       end
 
-      def pass_boundary_with_warning(boundary)
-        warn_of("The anchor '#{boundary}' at index #{expression.ts} "\
-                'only works at ASCII word boundaries in JavaScript.')
-        boundary
+      def convert_boundary
+        if context.es_2018_or_higher? && context.enable_u_option
+          BOUNDARY_EXPANSION
+        else
+          pass_boundary_with_warning
+        end
+      end
+
+      def convert_nonboundary
+        if context.es_2018_or_higher? && context.enable_u_option
+          NONBOUNDARY_EXPANSION
+        else
+          pass_boundary_with_warning
+        end
+      end
+
+      # This is an approximation to the word boundary behavior in Ruby, c.f.
+      # https://github.com/ruby/ruby/blob/08476c45/tool/enc-unicode.rb#L130
+      W                     = '\d\p{L}\p{M}\p{Pc}'
+      BOUNDARY_EXPANSION    = "(?:(?<=[#{W}])(?=[^#{W}]|$)|(?<=[^#{W}]|^)(?=[#{W}]))"
+      NONBOUNDARY_EXPANSION = "(?<=[#{W}])(?=[#{W}])"
+
+      def pass_boundary_with_warning
+        warn_of("The anchor '#{data}' at index #{expression.ts} only works "\
+                'at ASCII word boundaries with targets below ES2018".')
+        pass_through
       end
     end
   end

data/lib/js_regex/converter/assertion_converter.rb

@@ -14,13 +14,24 @@ class JsRegex
       def convert_data
         case subtype
         when :lookahead, :nlookahead
-          build_group(head: pass_through, capturing: false)
+          keep_as_is
+        when :lookbehind
+          return keep_as_is if context.es_2018_or_higher?
+
+          warn_of_unsupported_feature('lookbehind', min_target: Target::ES2018)
+          build_passive_group
         when :nlookbehind
-          warn_of_unsupported_feature('negative lookbehind assertion')
-        else # :lookbehind, ...
-          build_unsupported_group
+          return keep_as_is if context.es_2018_or_higher?
+
+          warn_of_unsupported_feature('negative lookbehind', min_target: Target::ES2018)
+        else
+          warn_of_unsupported_feature
         end
       end
+
+      def keep_as_is
+        build_group(head: pass_through, capturing: false)
+      end
     end
   end
 end

data/lib/js_regex/converter/backreference_converter.rb

@@ -10,16 +10,30 @@ class JsRegex
 
       def convert_data
         case subtype
-        when :name_ref, :number, :number_ref, :number_rel_ref then convert_ref
-        when :name_call, :number_call, :number_rel_call       then convert_call
+        when :name_ref then convert_name_ref
+        when :number, :number_ref, :number_rel_ref then convert_to_plain_num_ref
+        when :name_call, :number_call, :number_rel_call then convert_call
         else # name_recursion_ref, number_recursion_ref, ...
           warn_of_unsupported_feature
         end
       end
 
-      def convert_ref
-        position = context.new_capturing_group_position(target_position)
-        Node.new('\\', Node.new(position.to_s, type: :backref_num))
+      def convert_name_ref
+        if context.es_2018_or_higher?
+          # ES 2018+ supports named backrefs, but only the angled-bracket syntax
+          Node.new("\\k<#{expression.name}>", reference: new_position, type: :backref)
+        else
+          convert_to_plain_num_ref
+        end
+      end
+
+      def convert_to_plain_num_ref
+        position = new_position
+        Node.new("\\#{position}", reference: position, type: :backref)
+      end
+
+      def new_position
+        context.new_capturing_group_position(target_position)
       end
 
       def target_position
@@ -31,7 +45,10 @@ class JsRegex
           return warn_of_unsupported_feature('whole-pattern recursion')
         end
         context.increment_local_capturing_group_count
-        convert_expression(expression.referenced_expression.unquantified_clone)
+        target_copy = expression.referenced_expression.unquantified_clone
+        # avoid "Duplicate capture group name" error in JS
+        target_copy.token = :capture if target_copy.is?(:named, :group)
+        convert_expression(target_copy)
       end
     end
   end

data/lib/js_regex/converter/base.rb

@@ -51,10 +51,14 @@ class JsRegex
         Converter.convert(expression, context)
       end
 
-      def warn_of_unsupported_feature(description = nil)
+      def warn_of_unsupported_feature(description = nil, min_target: nil)
         description ||= "#{subtype} #{expression.type}".tr('_', ' ')
-        full_desc = "#{description} '#{expression}'"
-        warn_of("Dropped unsupported #{full_desc} at index #{expression.ts}")
+        full_text = "Dropped unsupported #{description} '#{expression}' "\
+                    "at index #{expression.ts}"
+        if min_target
+          full_text += " (requires at least `target: '#{min_target}'`)"
+        end
+        warn_of(full_text)
         drop
       end
 
@@ -68,11 +72,11 @@ class JsRegex
       alias drop_without_warning drop
 
       def wrap_in_backrefed_lookahead(content)
-        backref_num = context.capturing_group_count + 1
-        backref_num_node = Node.new(backref_num.to_s, type: :backref_num)
+        number = context.capturing_group_count + 1
+        backref_node = Node.new("\\#{number}", reference: number, type: :backref)
         context.increment_local_capturing_group_count
         # an empty passive group (?:) is appended as literal digits may follow
-        Node.new('(?=(', *content, '))\\', backref_num_node, '(?:)')
+        Node.new('(?=(', *content, '))', backref_node, '(?:)')
       end
     end
   end

data/lib/js_regex/converter/conditional_converter.rb

@@ -10,12 +10,12 @@ class JsRegex
 
       def convert_data
         case subtype
-        when :open then mark_conditional
+        when :open then mark_conditional_for_second_pass
         else warn_of_unsupported_feature
         end
       end
 
-      def mark_conditional
+      def mark_conditional_for_second_pass
         reference = expression.referenced_expression.number
         node = Node.new('(?:', reference: reference, type: :conditional)
         expression.branches.each do |branch|

data/lib/js_regex/converter/context.rb

@@ -11,12 +11,36 @@ class JsRegex
                   :in_atomic_group,
                   :warnings
 
-      def initialize(case_insensitive_root: false)
+      def initialize(case_insensitive_root: false, target: nil)
         self.added_capturing_groups_after_group = Hash.new(0)
         self.capturing_group_count = 0
         self.warnings = []
+        self.required_options_hash = {}
 
         self.case_insensitive_root = case_insensitive_root
+        self.target = target
+      end
+
+      # target context
+
+      def es_2015_or_higher?
+        target >= Target::ES2015
+      end
+
+      def es_2018_or_higher?
+        target >= Target::ES2018
+      end
+
+      # these methods allow appending options to the final Conversion output
+
+      def enable_u_option
+        return false unless es_2015_or_higher?
+
+        required_options_hash['u'] = true
+      end
+
+      def required_options
+        required_options_hash.keys
       end
 
       # group context
@@ -54,7 +78,9 @@ class JsRegex
 
       private
 
-      attr_accessor :added_capturing_groups_after_group
+      attr_accessor :added_capturing_groups_after_group,
+                    :required_options_hash,
+                    :target
 
       attr_writer :capturing_group_count,
                   :case_insensitive_root,

data/lib/js_regex/converter/escape_converter.rb

@@ -41,7 +41,7 @@ class JsRegex
         when :control, :meta_sequence
           unicode_escape_codepoint
         when :literal
-          LiteralConverter.convert_data(expression.char)
+          LiteralConverter.convert_data(expression.char, context)
         when *ESCAPES_SHARED_BY_RUBY_AND_JS
           pass_through
         when :bell, :escape, :octal
@@ -52,11 +52,19 @@ class JsRegex
       end
 
       def convert_codepoint_list
-        expression.chars.each_with_object(Node.new) do |char, node|
-          node << LiteralConverter.convert_data(Regexp.escape(char))
+        if context.enable_u_option
+          split_codepoint_list
+        else
+          expression.chars.each_with_object(Node.new) do |char, node|
+            node << LiteralConverter.convert_data(Regexp.escape(char), context)
+          end
         end
       end
 
+      def split_codepoint_list
+        expression.codepoints.map { |cp| "\\u{#{cp.to_s(16).upcase}}" }.join
+      end
+
       def unicode_escape_codepoint
         "\\u#{expression.codepoint.to_s(16).upcase.rjust(4, '0')}"
       end

data/lib/js_regex/converter/group_converter.rb

@@ -10,19 +10,30 @@ class JsRegex
 
       def convert_data
         case subtype
-        when :capture, :named then build_group
+        when :capture then build_group
+        when :named then build_named_group
         when :atomic then emulate_atomic_group
         when :comment then drop_without_warning
         when :options, :options_switch then build_options_group
         when :passive then build_passive_group
         when :absence then build_absence_group_if_simple
-        else build_unsupported_group
+        else warn_of_unsupported_feature
+        end
+      end
+
+      def build_named_group
+        if context.es_2018_or_higher?
+          # ES 2018+ supports named groups, but only the angled-bracket syntax
+          build_group(head: "(?<#{expression.name}>")
+        else
+          build_group
         end
       end
 
       def emulate_atomic_group
         if context.in_atomic_group
-          build_unsupported_group('nested atomic group')
+          warn_of_unsupported_feature('nested atomic group')
+          build_passive_group
         else
           context.start_atomic_group
           result = wrap_in_backrefed_lookahead(convert_subexpressions)
@@ -68,11 +79,6 @@ class JsRegex
         build_group(head: head, tail: tail, capturing: false)
       end
 
-      def build_unsupported_group(description = nil)
-        warn_of_unsupported_feature(description)
-        build_passive_group
-      end
-
       def build_group(opts = {})
         head = opts[:head] || '('
         tail = opts[:tail] || ')'

data/lib/js_regex/converter/keep_converter.rb

@@ -0,0 +1,24 @@
+require_relative 'base'
+
+class JsRegex
+  module Converter
+    #
+    # Template class implementation.
+    #
+    class KeepConverter < JsRegex::Converter::Base
+      private
+
+      def convert_data
+        if context.es_2018_or_higher?
+          if expression.level.zero?
+            Node.new(type: :keep_mark) # mark for conversion in SecondPass
+          else
+            warn_of_unsupported_feature('nested keep mark')
+          end
+        else
+          warn_of_unsupported_feature('keep mark', min_target: Target::ES2018)
+        end
+      end
+    end
+  end
+end

data/lib/js_regex/converter/literal_converter.rb

@@ -7,11 +7,15 @@ class JsRegex
     #
     class LiteralConverter < JsRegex::Converter::Base
       class << self
-        ASTRAL_PLANE_CODEPOINT_PATTERN = /[\u{10000}-\u{FFFFF}]/
+        ASTRAL_PLANE_CODEPOINT_PATTERN = /[\u{10000}-\u{10FFFF}]/
 
-        def convert_data(data)
+        def convert_data(data, context)
           if data =~ ASTRAL_PLANE_CODEPOINT_PATTERN
-            convert_astral_data(data)
+            if context.enable_u_option
+              escape_incompatible_bmp_literals(data)
+            else
+              convert_astral_data(data)
+            end
           else
             escape_incompatible_bmp_literals(data)
           end
@@ -41,7 +45,7 @@ class JsRegex
       private
 
       def convert_data
-        result = self.class.convert_data(data)
+        result = self.class.convert_data(data, context)
         if context.case_insensitive_root && !expression.case_insensitive?
           warn_of_unsupported_feature('nested case-sensitive literal')
         elsif !context.case_insensitive_root && expression.case_insensitive?

data/lib/js_regex/converter/meta_converter.rb

@@ -8,6 +8,16 @@ class JsRegex
     class MetaConverter < JsRegex::Converter::Base
       DOT_EXPANSION    = '(?:[\uD800-\uDBFF][\uDC00-\uDFFF]|[^\n\uD800-\uDFFF])'
       ML_DOT_EXPANSION = '(?:[\uD800-\uDBFF][\uDC00-\uDFFF]|[^\uD800-\uDFFF])'
+      # Possible improvements for dot conversion:
+      #
+      # In ES2015, the 'u' flag allows dots to match astral chars. Unfortunately
+      # the dot keeps matching lone surrogates even with this flag, so the use
+      # of an expansion is still necessary to get the same behavior as in Ruby.
+      #
+      # ES2018 has the dotall flag 's', but it is tricky to use in conversions.
+      # 's' activates matching of BOTH astral chars and "\n", whereas the dot in
+      # Ruby doesn't match "\n" by default, and even with the 'm' flag set on
+      # the root, subexps might still exclude "\n" like so: /.(?-m:.)./m
 
       private
 

data/lib/js_regex/converter/property_converter.rb

@@ -10,9 +10,24 @@ class JsRegex
     # codepoints matched by the property and build a set string from them.
     #
     class PropertyConverter < JsRegex::Converter::Base
+      # A map of normalized Ruby property names to names supported by ES2018+.
+      def self.map
+        @map ||= File.read("#{__dir__}/property_map.csv").scan(/(.+),(.+)/).to_h
+      end
+
       private
 
       def convert_data
+        if context.es_2018_or_higher? &&
+            (prop_name_in_js = self.class.map[subtype.to_s.tr('_', '')])
+          context.enable_u_option
+          "\\#{expression.negative? ? 'P' : 'p'}{#{prop_name_in_js}}"
+        else
+          build_character_set
+        end
+      end
+
+      def build_character_set
         content = CharacterSet.of_expression(expression)
 
         if expression.case_insensitive? && !context.case_insensitive_root

data/lib/js_regex/converter/property_map.csv

@@ -0,0 +1,171 @@
+# THIS FILE IS GENERATED BY $ rake build_prop_map - DO NOT EDIT
+ascii,ASCII
+asciihexdigit,ASCII_Hex_Digit
+adlam,Script=Adlam
+anatolianhieroglyphs,Script=Anatolian_Hieroglyphs
+armenian,Script=Armenian
+avestan,Script=Avestan
+bamum,Script=Bamum
+bassavah,Script=Bassa_Vah
+batak,Script=Batak
+bengali,Script=Bengali
+bhaiksuki,Script=Bhaiksuki
+bidicontrol,Bidi_Control
+bopomofo,Script=Bopomofo
+braille,Script=Braille
+buginese,Script=Buginese
+buhid,Script=Buhid
+carian,Script=Carian
+caucasianalbanian,Script=Caucasian_Albanian
+chakma,Script=Chakma
+cham,Script=Cham
+cherokee,Script=Cherokee
+chorasmian,Script=Chorasmian
+connectorpunctuation,Connector_Punctuation
+control,Control
+coptic,Script=Coptic
+cuneiform,Script=Cuneiform
+cypriot,Script=Cypriot
+cyrillic,Script=Cyrillic
+deprecated,Deprecated
+deseret,Script=Deseret
+devanagari,Script=Devanagari
+divesakuru,Script=Dives_Akuru
+dogra,Script=Dogra
+duployan,Script=Duployan
+egyptianhieroglyphs,Script=Egyptian_Hieroglyphs
+elbasan,Script=Elbasan
+elymaic,Script=Elymaic
+emojicomponent,Emoji_Component
+emojimodifier,Emoji_Modifier
+enclosingmark,Enclosing_Mark
+finalpunctuation,Final_Punctuation
+georgian,Script=Georgian
+gothic,Script=Gothic
+grantha,Script=Grantha
+greek,Script=Greek
+gujarati,Script=Gujarati
+gunjalagondi,Script=Gunjala_Gondi
+gurmukhi,Script=Gurmukhi
+hangul,Script=Hangul
+hanifirohingya,Script=Hanifi_Rohingya
+hanunoo,Script=Hanunoo
+hatran,Script=Hatran
+hebrew,Script=Hebrew
+hexdigit,Hex_Digit
+idsbinaryoperator,IDS_Binary_Operator
+idstrinaryoperator,IDS_Trinary_Operator
+imperialaramaic,Script=Imperial_Aramaic
+initialpunctuation,Initial_Punctuation
+inscriptionalpahlavi,Script=Inscriptional_Pahlavi
+inscriptionalparthian,Script=Inscriptional_Parthian
+javanese,Script=Javanese
+joincontrol,Join_Control
+kayahli,Script=Kayah_Li
+kharoshthi,Script=Kharoshthi
+khitansmallscript,Script=Khitan_Small_Script
+khmer,Script=Khmer
+khojki,Script=Khojki
+khudawadi,Script=Khudawadi
+lao,Script=Lao
+lepcha,Script=Lepcha
+letternumber,Letter_Number
+limbu,Script=Limbu
+lineseparator,Line_Separator
+lineara,Script=Linear_A
+linearb,Script=Linear_B
+lisu,Script=Lisu
+logicalorderexception,Logical_Order_Exception
+lycian,Script=Lycian
+lydian,Script=Lydian
+mahajani,Script=Mahajani
+makasar,Script=Makasar
+malayalam,Script=Malayalam
+mandaic,Script=Mandaic
+manichaean,Script=Manichaean
+marchen,Script=Marchen
+masaramgondi,Script=Masaram_Gondi
+math,Math
+mathsymbol,Math_Symbol
+medefaidrin,Script=Medefaidrin
+meeteimayek,Script=Meetei_Mayek
+mendekikakui,Script=Mende_Kikakui
+meroiticcursive,Script=Meroitic_Cursive
+meroitichieroglyphs,Script=Meroitic_Hieroglyphs
+miao,Script=Miao
+modi,Script=Modi
+mro,Script=Mro
+multani,Script=Multani
+myanmar,Script=Myanmar
+nabataean,Script=Nabataean
+nandinagari,Script=Nandinagari
+newtailue,Script=New_Tai_Lue
+newa,Script=Newa
+nko,Script=Nko
+noncharactercodepoint,Noncharacter_Code_Point
+nushu,Script=Nushu
+nyiakengpuachuehmong,Script=Nyiakeng_Puachue_Hmong
+ogham,Script=Ogham
+olchiki,Script=Ol_Chiki
+oldhungarian,Script=Old_Hungarian
+olditalic,Script=Old_Italic
+oldnortharabian,Script=Old_North_Arabian
+oldpermic,Script=Old_Permic
+oldpersian,Script=Old_Persian
+oldsogdian,Script=Old_Sogdian
+oldsoutharabian,Script=Old_South_Arabian
+oldturkic,Script=Old_Turkic
+oriya,Script=Oriya
+osage,Script=Osage
+osmanya,Script=Osmanya
+othernumber,Other_Number
+pahawhhmong,Script=Pahawh_Hmong
+palmyrene,Script=Palmyrene
+paragraphseparator,Paragraph_Separator
+patternsyntax,Pattern_Syntax
+patternwhitespace,Pattern_White_Space
+paucinhau,Script=Pau_Cin_Hau
+phagspa,Script=Phags_Pa
+phoenician,Script=Phoenician
+privateuse,Private_Use
+psalterpahlavi,Script=Psalter_Pahlavi
+quotationmark,Quotation_Mark
+radical,Radical
+regionalindicator,Regional_Indicator
+rejang,Script=Rejang
+runic,Script=Runic
+samaritan,Script=Samaritan
+saurashtra,Script=Saurashtra
+separator,Separator
+sharada,Script=Sharada
+shavian,Script=Shavian
+siddham,Script=Siddham
+signwriting,Script=SignWriting
+sinhala,Script=Sinhala
+sogdian,Script=Sogdian
+sorasompeng,Script=Sora_Sompeng
+soyombo,Script=Soyombo
+spaceseparator,Space_Separator
+sundanese,Script=Sundanese
+sylotinagri,Script=Syloti_Nagri
+syriac,Script=Syriac
+tagbanwa,Script=Tagbanwa
+taile,Script=Tai_Le
+taitham,Script=Tai_Tham
+taiviet,Script=Tai_Viet
+tamil,Script=Tamil
+tangut,Script=Tangut
+thaana,Script=Thaana
+thai,Script=Thai
+tibetan,Script=Tibetan
+tifinagh,Script=Tifinagh
+tirhuta,Script=Tirhuta
+titlecaseletter,Titlecase_Letter
+ugaritic,Script=Ugaritic
+vai,Script=Vai
+wancho,Script=Wancho
+warangciti,Script=Warang_Citi
+whitespace,White_Space
+yezidi,Script=Yezidi
+yi,Script=Yi
+zanabazarsquare,Script=Zanabazar_Square

data/lib/js_regex/converter/set_converter.rb

@@ -26,7 +26,12 @@ class JsRegex
           warn_of_unsupported_feature('nested case-sensitive set')
         end
 
-        content.to_s_with_surrogate_ranges
+        if context.es_2015_or_higher?
+          context.enable_u_option if content.astral_part?
+          content.to_s(format: 'es6', in_brackets: true)
+        else
+          content.to_s_with_surrogate_ranges
+        end
       end
 
       def directly_compatible?
@@ -41,8 +46,8 @@ class JsRegex
       def child_directly_compatible?(exp)
         case exp.type
         when :literal
-          # surrogate pair substitution needed if astral
-          exp.text.ord <= 0xFFFF
+          # surrogate pair substitution needed on ES2009 if astral
+          exp.text.ord <= 0xFFFF || context.enable_u_option
         when :set
           # conversion needed for nested sets, intersections
           exp.token.equal?(:range)

data/lib/js_regex/converter/type_converter.rb

@@ -6,9 +6,11 @@ class JsRegex
     # Template class implementation.
     #
     class TypeConverter < JsRegex::Converter::Base
-      HEX_EXPANSION       = '[0-9A-Fa-f]'
-      NONHEX_EXPANSION    = '[^0-9A-Fa-f]'
-      LINEBREAK_EXPANSION = '(?:\r\n|[\n\v\f\r\u0085\u2028\u2029])'
+      HEX_EXPANSION           = '[0-9A-Fa-f]'
+      NONHEX_EXPANSION        = '[^0-9A-Fa-f]'
+      ES2018_HEX_EXPANSION    = '\p{AHex}'
+      ES2018_NONHEX_EXPANSION = '\P{AHex}'
+      LINEBREAK_EXPANSION     = '(?:\r\n|[\n\v\f\r\u0085\u2028\u2029])'
 
       def self.directly_compatible?(expression)
         case expression.token
@@ -23,8 +25,8 @@ class JsRegex
 
       def convert_data
         case subtype
-        when :hex then HEX_EXPANSION
-        when :nonhex then NONHEX_EXPANSION
+        when :hex then hex_expansion
+        when :nonhex then nonhex_expansion
         when :linebreak then LINEBREAK_EXPANSION
         when :digit, :space, :word
           return pass_through if self.class.directly_compatible?(expression)
@@ -37,6 +39,22 @@ class JsRegex
         end
       end
 
+      def hex_expansion
+        if context.es_2018_or_higher? && context.enable_u_option
+          ES2018_HEX_EXPANSION
+        else
+          HEX_EXPANSION
+        end
+      end
+
+      def nonhex_expansion
+        if context.es_2018_or_higher? && context.enable_u_option
+          ES2018_NONHEX_EXPANSION
+        else
+          NONHEX_EXPANSION
+        end
+      end
+
       def negative_set_substitution
         # ::of_expression returns an inverted set for negative expressions,
         # so we need to un-invert before wrapping in [^ and ]. Kinda lame.

data/lib/js_regex/converter.rb

@@ -13,6 +13,7 @@ class JsRegex
       expression:  SubexpressionConverter,
       free_space:  FreespaceConverter,
       group:       GroupConverter,
+      keep:        KeepConverter,
       literal:     LiteralConverter,
       meta:        MetaConverter,
       nonproperty: PropertyConverter,

data/lib/js_regex/node.rb

@@ -9,10 +9,11 @@ class JsRegex
     attr_reader :children, :quantifier, :reference, :type
 
     TYPES = %i[
-      backref_num
+      backref
       captured_group
       conditional
       dropped
+      keep_mark
       plain
     ].freeze
 
@@ -46,7 +47,7 @@ class JsRegex
       case type
       when :dropped
         ''
-      when :backref_num, :captured_group, :plain
+      when :backref, :captured_group, :plain
         children.join << quantifier.to_s
       else
         raise TypeError.new(
@@ -59,6 +60,7 @@ class JsRegex
       self.children   = attrs.fetch(:children)   if attrs.key?(:children)
       self.quantifier = attrs.fetch(:quantifier) if attrs.key?(:quantifier)
       self.type       = attrs.fetch(:type)       if attrs.key?(:type)
+      self
     end
 
     private

data/lib/js_regex/second_pass.rb

@@ -6,12 +6,26 @@ class JsRegex
   module SecondPass
     class << self
       def call(tree)
+        substitute_root_level_keep_mark(tree)
         alternate_conditional_permutations(tree)
         tree
       end
 
       private
 
+      def substitute_root_level_keep_mark(tree)
+        keep_mark_index = nil
+        tree.children.each.with_index do |child, i|
+          break keep_mark_index = i if child.type == :keep_mark
+        end
+        return unless keep_mark_index
+
+        pre = tree.children[0...keep_mark_index]
+        post = tree.children[(keep_mark_index + 1)..-1]
+        lookbehind = Node.new('(?<=', *pre, ')')
+        tree.update(children: [lookbehind, *post])
+      end
+
       def alternate_conditional_permutations(tree)
         permutations = conditional_tree_permutations(tree)
         return if permutations.empty?
@@ -23,16 +37,16 @@ class JsRegex
       end
 
       def conditional_tree_permutations(tree)
-        all_conds = conditions(tree)
-        return [] if all_conds.empty?
+        conds = conditions(tree)
+        return [] if conds.empty?
 
         caps_per_branch = captured_group_count(tree)
 
-        condition_permutations(all_conds).map.with_index do |truthy_conds, i|
+        condition_permutations(conds).map.with_index do |truthy_conds, i|
           tree_permutation = tree.clone
           # find referenced groups and conditionals and make one-sided
           crawl(tree_permutation) do |node|
-            build_permutation(node, all_conds, truthy_conds, caps_per_branch, i)
+            build_permutation(node, conds, truthy_conds, caps_per_branch, i)
           end
         end
       end
@@ -63,16 +77,30 @@ class JsRegex
         end
       end
 
-      def build_permutation(node, all_conds, truthy_conds, caps_per_branch, i)
+      def build_permutation(node, conds, truthy_conds, caps_per_branch, i)
         truthy = truthy_conds.include?(node.reference)
 
-        if node.type.equal?(:captured_group) &&
-          all_conds.include?(node.reference)
-          adapt_referenced_group_to_permutation(node, truthy)
-        elsif node.type.equal?(:conditional)
-          adapt_conditional_to_permutation(node, truthy)
-        elsif node.type.equal?(:backref_num)
+        case node.type
+        when :backref
+          # We cannot use named groups or backrefs in the conditional expansion,
+          # their repetition would cause a "Duplicate capture group name" error in JS.
+          node.update(children: [
+            node.children.first.sub(/k<.*>/, node.reference.to_s)
+          ])
+          # backref numbers need to be incremented for subsequent "branches"
           adapt_backref_to_permutation(node, caps_per_branch, i)
+        when :captured_group
+          # Remove name, c.f. :backref handling.
+          node.update(children: [
+            node.children.first.sub(/\?<.*>/, ''),
+            *node.children[1..-1]
+          ])
+          # if the group is referenced by any condition, modulate its quantity
+          if conds.include?(node.reference)
+            adapt_referenced_group_to_permutation(node, truthy)
+          end
+        when :conditional
+          adapt_conditional_to_permutation(node, truthy)
         end
       end
 
@@ -91,8 +119,8 @@ class JsRegex
       end
 
       def adapt_backref_to_permutation(backref_node, caps_per_branch, i)
-        new_num = backref_node.children[0].to_i + caps_per_branch * i
-        backref_node.update(children: [new_num.to_s])
+        new_num = backref_node.reference + caps_per_branch * i
+        backref_node.update(children: ["\\#{new_num}"])
       end
 
       def min_quantify(node)

data/lib/js_regex/target.rb

@@ -0,0 +1,19 @@
+class JsRegex
+  module Target
+    ES2009 = 'ES2009'
+    ES2015 = 'ES2015'
+    ES2018 = 'ES2018'
+    SUPPORTED = [ES2009, ES2015, ES2018].freeze
+
+    def self.cast(arg)
+      return ES2009 if arg.nil?
+
+      normalized_arg = arg.to_s.upcase
+      return normalized_arg if SUPPORTED.include?(normalized_arg)
+
+      raise ArgumentError.new(
+        "Unknown target: #{arg.inspect}. Try one of #{SUPPORTED}."
+      ).extend(JsRegex::Error)
+    end
+  end
+end

data/lib/js_regex/version.rb

@@ -1,3 +1,3 @@
 class JsRegex
-  VERSION = '3.7.2'
+  VERSION = '3.8.0'
 end

data/lib/js_regex.rb

@@ -12,10 +12,10 @@ class JsRegex
   require_relative File.join('js_regex', 'version')
   require 'json'
 
-  attr_reader :source, :options, :warnings
+  attr_reader :source, :options, :warnings, :target
 
-  def initialize(ruby_regex, options: nil)
-    @source, @options, @warnings = Conversion.of(ruby_regex, options: options)
+  def initialize(ruby_regex, **kwargs)
+    @source, @options, @warnings, @target = Conversion.of(ruby_regex, **kwargs)
   end
 
   def to_h
@@ -30,8 +30,8 @@ class JsRegex
     "/#{source.empty? ? '(?:)' : source}/#{options}"
   end
 
-  def self.new!(ruby_regex, options: nil)
-    js_regex = new(ruby_regex, options: options)
+  def self.new!(ruby_regex, **kwargs)
+    js_regex = new(ruby_regex, **kwargs)
     if js_regex.warnings.any?
       raise StandardError.new(
         "Could not fully convert the given regex #{ruby_regex.inspect}:\n" +

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: js_regex
 version: !ruby/object:Gem::Version
-  version: 3.7.2
+  version: 3.8.0
 platform: ruby
 authors:
 - Janosch Müller
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-05-27 00:00:00.000000000 Z
+date: 2022-09-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: character_set
@@ -72,9 +72,11 @@ files:
 - lib/js_regex/converter/escape_converter.rb
 - lib/js_regex/converter/freespace_converter.rb
 - lib/js_regex/converter/group_converter.rb
+- lib/js_regex/converter/keep_converter.rb
 - lib/js_regex/converter/literal_converter.rb
 - lib/js_regex/converter/meta_converter.rb
 - lib/js_regex/converter/property_converter.rb
+- lib/js_regex/converter/property_map.csv
 - lib/js_regex/converter/set_converter.rb
 - lib/js_regex/converter/subexpression_converter.rb
 - lib/js_regex/converter/type_converter.rb
@@ -82,6 +84,7 @@ files:
 - lib/js_regex/error.rb
 - lib/js_regex/node.rb
 - lib/js_regex/second_pass.rb
+- lib/js_regex/target.rb
 - lib/js_regex/version.rb
 homepage: https://github.com/jaynetics/js_regex
 licenses: